On May 15, 2005, at 10:24 AM, Tim Bray wrote:
On May 14, 2005, at 11:02 AM, A. Pagaltzis wrote:
The "atom:content" element either contains or links to the full content of the entry. An "atom:entry" containing an "atom:content" element MUST be a complete representation of the entry.
-1
What does "complete" mean? This is untestable and semantically fuzzy. I think anything stronger than "is intended to contain the full content" is just not workable. -Tim
Tim, that is exactly the point.
When parsing RSS <description>, there is no way to know if what is there is complete or an extract. Atom rightly expresses these as semantically distinct entities. There isn't a good way to distinguish them algorithmically, so we really need explicit language in the spec making this clear.
It needs human validation, as it can't be done by code, but it needs to be in the spec so we can point implementers to it, or better, that they see it and realise why there are 2 distinct elements for these purposes.
Putting abbreviated forms in both is the semantic pollution; if an entry is brief enough to be unsummarizable, having the same in both summary and content is perhaps legitimate, but if making the 'not the same' requirement explicit requires that very short entries have no summary, that works and increases feed consistency.
The difficulty comes when people view the feed content as something distinct from the content of the entry itself; in that case we want them to use atom:summary rather than atom:content. Just saying 'they can't be the same' is insufficient; the reason they are distinct needs to be there in the specification too.
My initial language may be clumsy or overdone, but this requires a MUST rather than a SHOULD if the content/summary elements are to remain useful and distinct.
